Skip to content

AI inference

Build with AI.
Work with AI.

One OpenAI-compatible endpoint for the applications you ship and the coding tools you use. Chat, embeddings, image and video generation and model choice sit behind the same organization key — and every call draws down the same prepaid credit as the apps and databases Maxlayer already runs for you.

Start using AI Browse live models
openai.ts 
import OpenAI from "openai";const client = new OpenAI({  apiKey: process.env.MAXLAYER_API_KEY,  baseURL: "https://inference.maxlayer.cloud/v1",});const completion = await client.chat.completions.create({  model: process.env.MAXLAYER_MODEL_ID,  messages: [{ role: "user", content: "Explain this pull request." }],});completion ready

Interface

OpenAI-compatible

The API shape your tools already expect.

Capabilities

Chat, embeddings, images, video

Answer a question, index a repository or a picture library, or generate a still or a clip — from one endpoint.

Billing

One organization balance

Per-call usage, alongside your apps and databases.

Where it fits

The endpoint does not ask you to pick a side

Use it in the product you are building, in the tool helping you build it, or in both. The setup stays one base URL and one organization key.

Your application code

Use the OpenAI SDKs, LangChain, LlamaIndex, or the Vercel AI SDK. Change the base URL, keep the client you already wrote.

Your coding tools

Cursor, Continue, Cline, Aider, and Zed take the same endpoint. Use your organization key and the model id from the catalogue.

No second account

From credit to completion

Inference belongs to the same organization as the rest of your platform. There is no separate vendor account, payment method, or key to rotate.

  1. 01

    Create an organization key

    The same organization API key that automates your apps authorizes inference. A browser session token cannot spend your balance.

  2. 02

    Choose a listed model

    The public catalogue is the source of truth for models, capabilities, and what each one charges — token rates, or a price per image or per second.

  3. 03

    Point your client at Maxlayer

    Chat completions, embeddings, images and video use the OpenAI request and response shapes, including streamed chat responses.

The other half

The app making these calls can run here too

Inference is one face of the platform. The service behind your chat box, the database it reads, and the record of every deploy are the others — same organization, same key, same balance.

Run the service

Deploy the API behind your chat box from a repository or an image. Automatic builds, a hostname, a certificate, and the port you choose.

Store what it reads

PostgreSQL, MySQL, MariaDB, MongoDB or Redis, private to your project, with the connection string written into your app as a secret.

Pay from one balance

Apps and databases meter by the hour and cap monthly; inference meters per call. One prepaid credit, one place to read it.

Prices that do not go stale

The catalogue is public. The numbers are current.

Models can be listed, repriced, or removed without a marketing-site deploy. Rather than leave a copied price table to drift, the public catalogue reports the live selection and what each model charges — a token rate card where there is one, and the provider's own price per image, per second, or per kind of input where there is not — before you make a request.

Open the live model catalogue

Ready when your key is

An endpoint your stack already understands.

Create an organization API key, choose a model, and point your client at the base URL. That is the integration.

Get started Read the API guide